Comparing datapoints expressed as proportions My stats knowledge isn't very advanced, so it's been difficult to find any resources for the exact kind of comparison I want to do. As a summary:
I'm measuring insect development (i.e. how many bugs develop to adulthood from eggs). I have two groups (treatment and control) and within each group I have 5 replicates (separate containers). So the data I get from each experiment is a set of 5 control proportions and 5 treatment proportions.
Here is some sample data (number developing to adulthood/total number at start):
Control   Treatment
82/90     26/82
80/95     30/90
90/118    11/76
102/110   41/101
76/98     12/99

At first, a chi square test looked like the right way to go, but I'm unsure about lumping all my proportions together for such a test. Is there a better way to go about this?
 A: I would use a binomial generalised linear model, for which the data are perhaps more easily presented as something like
  +----------------------------+
  |     which   adults   total |
  |----------------------------|
  |   Control       82      90 |
  |   Control       80      95 |
  |   Control       90     118 |
  |   Control      102     110 |
  |   Control       76      98 |
  | Treatment       26      82 |
  | Treatment       30      90 |
  | Treatment       11      76 |
  | Treatment       41     101 |
  | Treatment       12      99 |
  +----------------------------+

In Stata, that could be glm adults i.which, f(binomial total) on the understanding that which has integer values underneath its text labels. The implied link is logit, so this could also be described as logistic regression. The logit model converges on the sample proportions of survival of about 0.842 and 0.268 for control and treatment, with suitably overwhelming figures of merit. I am confident that any decent statistical software should support this standard model, although that is a definition of decent.
